def main():
arg = argparser().parse_args()
query = "select i-id where p-id = {}".format(arg.pid)
tsdb_results = delphin.tsdb_query(query, arg.profile)
gold_iids = set(int(x) for x in tsdb_results.split())
all_iids = delphin.get_profile_ids(arg.profile)
predicted_iids = get_predicted_iids(arg)
print evaluate(predicted_iids, gold_iids, len(all_iids))
def index(profiles, treebank, in_grammar):
stats_dict = defaultdict(delphin.TypeStats)
trees = 0
failures = []
for path in profiles:
grammar = in_grammar
items_seen = set()
print("processing {}".format(path))
profile = os.path.basename(path)
if profile in ERG_SPEECH_PROFILES:
alias = grammar.alias+'-speech'
grammar = gram.get_grammar(alias)
try:
# for treebanked profiles:
out = delphin.tsdb_query('select i-id derivation where t-active > 0', path)
# for non-treebanked profiles
# out = delphin.tsdb_query('select i-id derivation where readings > 0', path)
except delphin.TsdbError as e:
sys.stderr.write(str(e)+'\n')
continue
if out == '':
continue
results = out.strip().split('\n')
for result in results:
iid, derivation = result.split(' | ')
if iid in items_seen or iid in BLACKLIST:
continue
try:
counts = get_types(derivation, grammar)
for name, count in counts.items():
stats_dict[name].update(count)
except delphin.AceError as e:
e.other_data.append(iid)
e.other_data.append(path)
failures.append(e)
sys.stderr.write(str(e) + '\n')
else:
items_seen.add(iid)
trees += 1
print(trees, iid)
print("Processed {} trees".format(trees))
num_failures = len(failures)
if num_failures > 0:
print("Failed to reconstruct {} trees".format(num_failures))
print("See type-stats-errors.txt for details.")
with open('type-stats-errors.txt', 'w') as f:
errors_str = '\n'.join(str(e) for e in failures)
f.write(errors_str.encode('utf8')+'\n\n')
treebank_str = treebank.replace(' ', '_')
filename = '{}--{}--{}.pickle'.format(grammar.alias, treebank_str, trees)
with open(filename, 'wb') as f:
pickle.dump(stats_dict, f)